1. Field of the Invention
The present invention relates generally to a method and apparatus for automatically configuring a network. More particularly, the present invention relates to a dynamically configurable network architecture containing networked office appliances (NOAs).
2. Background
Generally, before a device can operate in a networked environment, the device needs to be appropriately configured. The specific network configuration required for each device may vary depending upon factors such as network topology and device functionality. On a Transmission Control Protocol/Internet Protocol (TCP/IP) network, each device or host is assigned a unique identifier known as an IP address. IP addresses usually take the form of a 32-bit numeric address written as four numbers separated by periods, but may differ depending upon the particular network topology. More information regarding IP address formatting can be found in Request For Comments (RFC) 1700, “Assigned Numbers”, October 1994.
Prior to establishing communication with a host across a TCP/IP network, the individual initiating the communication determines the IP address of the destination host much like telephone numbers are used when placing telephone calls. Individuals do not always know a recipient's IP address, and unfortunately there is no simple way of ascertaining it. Fortunately, however, mechanisms such as domain names have been developed which allow for simplified network communication.
Domain names were introduced as a less complicated means for users to communicate with remote hosts. Background information and specifications regarding domain names can be found in RFC 1034, “Domain Names—Concepts and Facilities”, November 1987, and a companion RFC 1035, “Domain Names—Implementation and Specification”, November 1987. In its simplest form, a domain name is nothing more than a human-readable text representation of one or more unique numeric IP addresses. By using domain names, users are not required to memorize numerous awkward IP addresses in order to communicate with hosts, rather they need only remember the appropriate host's domain name. Furthermore, most domain names reflect some variation of their corresponding hosts' names, and thus act as mnemonics for the user. Because the Internet is based upon IP addressing and not domain names, however, network devices replace every host domain name with the host's corresponding IP address prior to initiating communication. This replacement process is facilitated by what is known as a domain name service (DNS).
FIG. 1 illustrates a conventional DNS configuration according to the prior art. Referring to FIG. 1, client 100 transmits the domain name 102 of the host 115 that client 100 wishes to contact over network 101 to DNS server 120. DNS server 120 performs a lookup function in its database 118 and retrieves a corresponding IP address 103 to return to the client 100. Client 100 is then free to establish communication with host 115 using the host's corresponding IP address 103.
It is not uncommon for a single host domain name to be linked to multiple IP addresses within a DNS server. If one or more of a host's IP addresses are changed or are removed from the network, the corresponding DNS entries also need to be changed or removed from the DNS database. One limitation of DNS has historically been its inability to automatically update such address changes. Currently, it is common for DNS updates to be performed manually, often requiring time consuming, meticulous precision. Such manual updates, however, are not limited to just DNS databases. Essential configuration information such as IP addresses have traditionally been assigned to hosts manually as well.
Manual IP address assignments are most often performed by an experienced individual such as a network administrator. The network administrator assigns each host an IP address chosen from a block of addresses known by the administrator to be available. If the administrator were to mistype the IP address when configuring the host, it is likely that the host would not function correctly. Similarly, if the administrator were to assign a previously allocated IP address to a host, communication errors would likely occur due to the conflicting IP addresses. Additional TCP/IP configuration information other than IP addresses is often manually entered as well. Often, the TCP/IP configuration process requires an administrator to visit each host individually to enter the applicable information. Every time a host configuration is subsequently changed, the administrator must also visit the host to perform the update. On large networks, this practice of manually updating configurations can be extremely time consuming.
Dynamic IP address allocation attempts to solve some of the problems created by manual host configuration. Different implementations of dynamic IP address allocation have been proposed over time, but one common protocol used today is the Dynamic Host Configuration Protocol (DHCP). For more information on DHCP, see RFC 2131, “Dynamic Host Configuration Protocol”, March 1997. DHCP is based on the Bootstrap Protocol (BOOTP), but adds the capability of automatic allocation of reusable network addresses and additional configuration options. For more information on BOOTP, see RFC 951, “Bootstrap Protocol. (BOOTP)”, September 1985. DHCP provides a framework for passing configuration information to hosts on a TCP/IP network at boot time. With DHCP, a network administrator does not need to visit each host individually to configure or modify a host's configuration. A configuration may include the host IP address and other TCP/IP option settings such as the definitions of domain name servers, default gateways and subnet masks. Some DHCP server implementations allow for the use of option sets, which allow administrators to assign common settings to particular options. When an administrator makes a change to an option set, all client configurations employing that option set will receive the updated information. In this manner, central administration is made easier.
FIG. 1 illustrates a conventional DHCP configuration according to prior art. Referring to FIG. 1, upon commencing its bootstrap routine, a DHCP client 100 sends out a DHCP Discover broadcast 105 across network 101 looking for a DHCP server 110 or 111 that can return settings to client 100. Both operational DHCP servers 110 and 111 on network 101 receive the DHCP Discover broadcast 105 from client 100 and determine if they can provide configuration information for that particular client 100. If the DHCP servers 110 and 111 have a configuration for the requesting client 100, they send a DHCP Offer 106 to the DHCP client 100 over network 101. DHCP client 100 analyzes all of the DHCP Offers 106 it has received from DHCP Servers 110 and 111, selects one of the servers, and sends back a DHCP Request 107 over network 101 to the server it chooses, such as DHCP server 110. DHCP server 110 issues a DHCP Acknowledgment 108 to client 100, reserves an IP address, and subsequently delivers the configuration information to client 100 over network 101.
One current problem with networks utilizing DHCP in conjunction with DNS involves the lack of communication between the two systems. As IP addresses are dynamically allocated to hosts, corresponding domain names may also be allocated. In a network where a host's IP address changes frequently, so too will its domain name. Due to the randomness involved in such address/domain name assignments, it is crucial that such assignments be reflected in a DNS. Since DNS updates are often completed manually, it is unlikely, however, that the DNS will reflect the most recent host address/domain name information available. If the DNS is not kept up to date with the most recent host address/domain name assignments, communication between hosts may become increasingly difficult. Even if a DNS database was kept up to date with the most recently available network address/domain name information, it would still be advantageous to identify a single host by a single fixed domain name. With a single fixed domain name, any particular host would always be reachable via that domain name, regardless of what the host's dynamically assigned IP address is.
It is therefore desirable to have a system that is able to dynamically allocate IP addresses to a host, while statically assigning a single domain name to that host. Such a system would perform necessary DNS updates automatically, requiring minimal human interaction with the system, thereby minimizing communication errors.
Once a device is configured, and connected to a network, the network administrator is generally responsible for updating user and group information and allocating access privileges to that device. As the size of the network, number of connected devices, and number of users grows, the process of updating user and group information and granting access privileges can become a formidable task. Often in large office environments, user and group information updates are not considered critical to the functionality of the network and may therefore be assigned a lower priority than other, more urgent system concerns. In small office environments designated network administrators may not even exist, leaving all of the network configurations and administration to be completed by untrained individuals. Such a practice may not only affect productivity, but may also jeopardize the functionality of the network.
It is therefore desirable to have a network device that provides easy, comfortable, and appliance-like automatic configuration features to users. Such a device should be capable of automatically configuring itself for network operation when placed in a network environment that lacks a designated administrator, and at the same time, the device should provide interoperability with preexisting networked equipment when placed in an administered network environment.